Skip to content

Remove TensorFlow dependency from multimodal image preprocessing #548

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
KumarADITHYA123 wants to merge 1 commit into
base: main
Choose a base branch
from
Open

Remove TensorFlow dependency from multimodal image preprocessing #548

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
12 changes: 7 additions & 5 deletions gemma/multimodal/image.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -18,12 +18,12 @@
from collections.abc import Sequence
import einops
from etils import epath
import io
import jax
from jax import numpy as jnp
from kauldron import typing
import numpy as np
from PIL import Image
import tensorflow as tf

_IMAGE_MEAN = (127.5,) * 3
_IMAGE_STD = (127.5,) * 3
Expand Down Expand Up @@ -70,10 +70,12 @@ def pre_process_image(
The pre-processed image.
"""
# all inputs are expected to have been jpeg compressed.
# TODO(eyvinec): we should remove tf dependency.
image = jnp.asarray(
tf.image.decode_jpeg(tf.io.encode_jpeg(image), channels=3)
)
# Perform JPEG encode/decode to normalize image (matches original behavior)
pil_image = Image.fromarray(np.uint8(image))
buffer = io.BytesIO()
pil_image.save(buffer, format="JPEG")
buffer.seek(0)
image = jnp.asarray(np.array(Image.open(buffer).convert("RGB")))
image = jax.image.resize(
image,
shape=(image_height, image_width, 3),
Expand Down